OpenAI’s GPT-6 Cyber found 81 Firefox flaws earlier AI testing missed

OpenAI's GPT-6 Cyber identified 81 high-severity Firefox vulnerabilities and 948 Linux kernel flaws that conventional security testing methods had previously missed, highlighting a significant gap in traditional vulnerability research approaches. This matters because unknown vulnerabilities represent an invisible attack surface — threat actors with access to similar AI capabilities could discover and exploit these flaws before defenders are even aware they exist. The fact that GPT-6 Cyber also demonstrated working exploit development for a critical Brotli vulnerability underscores that AI is now accelerating the arms race between attackers and defenders. Organizations relying solely on periodic manual audits or legacy scanning tools are increasingly outpaced by AI-augmented threat actors. This incident signals that vulnerability management programs must evolve to incorporate AI-assisted testing to remain effective.